Research on Seismic Performance of Bridge Piers Based on Machine Learning

Abstract

The shear capacity of the bridge pier component is an essential condition that directly affects the seismic performance of the pier and thus the earthquake resistance of the bridge. At present, 61.3 per cent of the short pier specimens in the database have a shear span ratio of less than 2.0, and 68.5 per cent of the specimens with a high axial compression ratio are greater than 0.4. There is no experimental data on irregular-shaped bridge piers of long-span bridges in areas with high-intensity earthquakes, and no experimental data on shear failure in the plastic hinge region under cyclic loading from strong earthquakes has been obtained. Generally speaking, the accuracy of the old formulas is lower than that of machine learning models.